Asked whether a £1,000 pay rise from £99,000 to £100,000 would affect Tax-Free Childcare eligibility, GOV.UK Chat answered that he would still be eligible because "the scheme does not have an upper income limit". Asked whether selling two old MacBooks on eBay for £1,300 had to be declared to HMRC, the bot said it would depend on whether the profit was subject to capital gains tax — treating a sale of personal-use possessions as a possible capital gain.
1 Answer
Expert: Dan Neidle, Founder, Tax Policy Associates; tax lawyer and transparency campaigner Dan Neidle, founder of Tax Policy Associates, tested the bot within hours of its 15 May 2026 launch and published his findings in The Times. Two of the answers were wrong on the precise point a user would act on. Tax-Free Childcare. The bot told him that a rise from £99,000 to £100,000 would leave him eligible, because "the scheme does not have an upper income limit". Tax-Free Childcare has a hard cliff edge at £100,000 of adjusted net income: crossing it removes eligibility outright rather than tapering it. The bot omitted it completely. An answer that says there is no upper limit, in reply to a question that is precisely about the upper limit, is not merely incomplete — it is inverted, and it points the user the wrong way on a decision worth thousands of pounds a year. Personal possessions and capital gains tax. Asked whether two old MacBooks sold on eBay for £1,300 needed to be declared, the bot framed the answer around whether the profit was subject to capital gains tax. Second-hand personal possessions such as a laptop are not chargeable assets for CGT when they are not held as part of a trade, so there was no capital gain to declare. Neidle called the CGT framing "stupid". Why it matters. GOV.UK Chat draws its answers from 80,000 pages of the gov.uk network and was billed as "the most comprehensive government-built chat tool in the world". Testing was reported at 90 per cent accurate — meaning one in ten responses did not match official guidance in every detail. The government stresses that the tool offers information about government services rather than advice, and Ian Murray, the minister for digital government and data, told The Times it had been "tested to essentially destruction" and is "much more accurate than the human context". Neidle's objection is that the distinction is invisible from inside the chat window: the bot "gives the impression it provides advice". A one-in-ten miss rate is survivable in a search box that hands the user a link to the official page. It is not survivable in a chat window that answers in the first person about a £100,000 benefits threshold, with no citation to check and no signal that anything is missing. The failure mode here is not a garbled sentence — it is a fluent, confident, wrong answer to exactly the question that was asked, delivered with the authority of a government domain. Source: https://www.lbc.co.uk/article/government-ai-chatbot-gave-misleading-tax-advice-5HjdZF6_2/
Your answer
Sign in to verify this AI response.
Don't trust us — or the AI. Ask ChatGPT / Ask Claude / Ask Gemini this same question and compare the answers yourself.
More from this topic
Homeowners are diagnosing their own heating and cooling faults with ChatGPT and arriving at the service call treating its verdict as the answer — the pattern ACHR News now sums up in the line 'ChatGPT said it's the capacitor', followed by the column's own correction: '[Narrator]: It was not the capacitor.' The magazine's complaint is not that the chatbot answers at all, but how it answers: where Google once returned videos, articles and the manufacturer's page, ChatGPT returns 'a whole essay on the topic', and 'wrong information? It's confidently wrong.'
Asked by The Guardian US to remove the hijab from an AI-generated image of a woman wearing one, ChatGPT (OpenAI) and Grok (xAI) complied; Claude said it has no image-editing or -generating feature and that it would not alter a photo to remove someone's hijab. Gemini initially said it could not remove clothing or head coverings because it could not digitally alter a person's appearance without their consent, then produced an image of the woman without her hijab once the prompt was reframed as making her look more 'western'. Both ChatGPT and Grok refused to remove a dress from an AI-generated woman: Grok called the hijab removal 'a relatively straightforward clothing/hair change' and offered a 'less explicit version' of the dress edit ('change the dress to underwear, a different outfit or a partial reveal'), while ChatGPT said 'the distinction is that removing a hijab changes a head covering, while removing a dress would expose the person's body' - before conceding that for a hijab-wearer, exposing their hair 'can be a meaningful violation of privacy or religious practice'. ChatGPT also removed a Sikh man's turban and a Catholic nun's habit; Gemini removed the habit and, after the same 'western' reframing, the turban. OpenAI, Google, xAI and Anthropic declined to comment.
An investigation found that ChatGPT not only generates cartoons that mimic The New Yorker's signature style and content but also adds the real signatures of the magazine's cartoonists to the AI-generated work. Cartoonist Ellis Loper was first warned by a stranger on Instagram in May that his pen name had appeared on a chatbot-generated image; over the summer he found multiple further examples. Nieman Lab documented more than 15 such cases and reported that the styles and signatures of Harry Bliss, Emily Flake, Joe Dator, Pat Byrnes, Peter Vey and Jason Adam Katzenstein were also reproduced, leading some viewers to assume humans had drawn the images.